The 1 sampling delay in PCM2906

Investigating as to why SPDIF is much better than USB interface on the D10. I started some additional investigation. Using a software waveform generator I generate square waves and measure it on the scope this is what I found.

This is a 1Khz Square wave using SPDIF. The left and right signal is place close together on a dual trace OSC-Scope


The wave form is mono and I adjusted the amplitiude on the OSC-Scope so that you could see the waveform side by side. The ringing (overshoot and undershoot) is normal.

Using USB and even trying with different computer this is what I get which is NOT normal.


As you could see highlighted by the 2 red arrow the signal is not in phase. It is about 4 degree out of phase, Or about one sample (sampling frequency) off. I could offset the phase in the signal generator so that is how I calculate the difference in Phase.
What this means is that one side of the sound is coming out slightly faster then the other. The Left and Right signal is suppose to be Mono or equivalent to Mono.

I will try to investigate more. Sorry for another troubling find.

Found an Errata from Texas Instrument.
http://focus.ti.com/lit/er/slaz036a/slaz036a.pdf

An extract from the Errata

Quote:
2.2 Inter-Channel Phase Difference
2.3 Exceeding Max-Packet Size
Problems, Restrictions, Workaround Analog, Embedded Processing, Semiconductor Company, Texas Instruments
Problems
The PCM290X has delay of 1 sample at each channels as follows:
· Up-stream data from ADC, S/PDIF in : Rch data delay (1 sample)
· Down-stream data to DAC, S/PDIF out : Lch data delay (1 sample)
Restrictions
Recording or playback of general audio or S/PDIF In/Out of Linear PCM, makes minimal difference to
the listener. However, the critical application for the inter-channel phase difference, data transfer
between USB and S/PDIF In/Out, can not be used.
Workaround
If the inter-channel difference is not allowed, the user must solve this problem in the application
software. It is not possible to solve this problem by any modification of external circuit.
TI is considering corrections to solve the problem, and will notify the user when a solution is
determined.
Take care to note the problems listed in this document when considering the use of these products for
new projects.

There is an upgraded part 2906B
And according to the article, publish in Jan 2009 at
http://focus.ti.com/lit/ml/slel056/slel056.pdf

Quote:
c. Fix of one-sample interchannel phase error in recording and playback.

If this is the case, Most of the existing product in the market would be affected as PCM290x is widely used.
EDIT: I wounder if the PCM270x are affected need to investigate

Using a USB to SPDIF is unlikely to solve the problem as the device is most likely using a PCM290x device. Unless it is using the upgraded version.

I need to catch some sleep.

I am really not expecting this.

Advertisements

5 responses to “The 1 sampling delay in PCM2906

  1. Cool site, love the info.

  2. Excellent site, keep up the good work

  3. Pingback: iBasso D10 USB vs. Optical in? - Page 2 - Head-Fi: Covering Headphones, Earphones and Portable Audio

  4. Hi,

    Has Ibasso announced a new version of the D10 with a working component replacing the PCM2906 ?

    • From what I understand iBasso is not changing the PCM2906 to the newer PCM2906B.

      A friend of mine is interested in buying a D10 with a newer PCM2906B but that is not available. Maybe iBasso will change their mind at a later date, Hopefully they do.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s